Major corporations are rapidly scaling back artificial intelligence deployments after discovering that operational costs far exceed initial projections. Uber reportedly exhausted its entire 2026 AI budget by April due to expensive engineering API calls. Amazon, Walmart, Cisco, and Meta have similarly curtailed tool usage as expenses spiral beyond forecasted limits.
AI inference now consumes approximately 85% of total enterprise AI budgets according to the FinOps Foundation. The prevailing per-token pricing model means every automated workflow significantly impacts financial reserves. Numerous firms report annual allocations vanishing within one to three months, with costs doubling or tripling year over year.
Executives are responding with strict rationing strategies including hard spending caps and prioritizing applications with proven returns on investment. Industry leaders describe current per-token cost increases as unsustainable alongside each new model generation.
This fiscal tightening carries significant implications for technology investors. Tokens tied to decentralized AI infrastructure may see renewed interest as enterprises seek alternatives to expensive centralized providers. Companies offering model optimization and efficient inference are now positioned to capture market share from cost-conscious corporate buyers.